Basic Science and Pathogenesis.

Background: Recent studies suggest genome-wide-association-studies (GWAS) loci confer their effects on microglia in late-onset Alzheimer's disease (LOAD) brains. Relatively fewer studies have investigated the effects of other genome-wide significant loci (p<5e) using human neurons.

Method: GWAS itself cannot directly identify causal variant-(effector)gene-pairs as GWAS only reports the sentinel variant at a given locus. To map the physical interactions between GWAS-implicated putative causal non-coding variants and their corresponding effector genes in LOAD, we leveraged the post-GWAS analytical framework, INFERNO, coupled with the largest harmonized functional genomics data catalog to date, FILER. Our goal was to unravel the effects of all 75 genome-wide significant genetic loci reported in Bellenguez et al in human brain cell types. We searched for plausible evidence of causality for LOAD candidate variants in any brain-specific variant-gene pair through a total of 3,873 publicly curated datasets (123 enhancer, 3,153 epigenetics, 105 QTL) across 12 resources. We then validated the top variant-gene pairs using high-resolution promoter-focused Capture C, ATAC-seq and RNA-seq datasets from several neuronal cell types and performed CRISPR interference.

Result: The computational variant-gene analyses identified a novel candidate causal SNP, rs74504435 (r2 = 0.94 to sentinel SNP rs76928645) at the SEC61G locus, located between SEC61G/EGFR. With a CADD score of 15, it is predicted as "regulatory" by both ENSEMBL and regulomeDB2. rs74504435 is found in ChromHMM and EpiMAP brain enhancers, and perturbs the transcriptional factor OCT6, which controls Schwann cell myelination. The minor allele T is protective to LOAD and downregulates EGFR in GTEx brain cortex (p = 2.3e), CommonMind (FDR p = 5.8e), and ROSMAP DLPFC (FDR p = 8.6e) datasets with the same effect direction. In our Capture C and ATAC-seq datasets, the SNP resides in open chromatin and directly contacts EGFR in human iPSC-derived cortical neurons and ES-derived hypothalamic neurons. We further validated a regulatory role for the enhancer region containing rs74504435 on EGFR using CRISPR interference in human microglial cells.

Conclusion: Together, we identified a novel LOAD-associated neuronal enhancer that drives EGFR gene expression. Further research is needed to investigate if EGFR inhibitors approved as anti-cancer drugs can serve as potential candidates for repurposing as LOAD therapeutics.